红外技术, 2018, 40 (2): 139, 网络出版: 2018-03-21
基于BMA滤波器和边缘的红外与可见光图像融合
Infrared and Visible Image Fusion Based on BMA Filter and Edge
红外图像 可见光图像 多尺度分解 BMA 滤波器 显著性提取 边缘权值映射 infrared images visible images multi-scale decomposition BMA filter saliency extraction edge weight map
摘要
目前基于多尺度分解的图像融合算法存在以下问题:1)多尺度分解时,图像边缘被平滑;2)融合结果中红外显著区域的对比度降低;3)小尺度细节受到抑制,在融合图像中显示不清晰。为解决上述问题,本文提出了一种基于BMA(Bayesian model averaging)滤波器和边缘的图像融合算法。首先,利用BMA 滤波器分别对红外与可见光图像进行多尺度分解;其次,分别利用显著性提取和边缘权值映射算法,计算各基层和细节层的融合权值矩阵;最后通过图像重构获得融合图像。实验证明,该融合算法优于传统的图像融合算法。
Abstract
Image fusion methods based on multi-scale decomposition have some problems as follows: 1) the edge is blurred in the multi-scale decomposition process; 2) contrast of the infrared salient region in the fused image is reduced; 3) small-scale details are suppressed in the fusion process, which looks dim. To address these problems, a new method of infrared and visible image fusion based on the Bayesian model averaging(BMA) filter and edge is proposed in this paper. First, both infrared and visible images are decomposed by the BMA filter; second, a saliency extraction method and an edge weight map are used to compute the weight matrix of base layers and detail layers, respectively; finally, the fused image can be obtained by image reconstruction. Experiments demonstrate that the proposed method outperforms the traditional fusion methods.
李博博, 马泳, 张晓晔, 樊凡. 基于BMA滤波器和边缘的红外与可见光图像融合[J]. 红外技术, 2018, 40(2): 139. LI Bobo, MA Yong, ZHANG Xiaoye, FAN Fan. Infrared and Visible Image Fusion Based on BMA Filter and Edge[J]. Infrared Technology, 2018, 40(2): 139.